canSAR chemistry registration and standardization pipeline

Area of Science:

• Medicinal Chemistry
• Drug Discovery
• Bioinformatics

Background:

• Integrating medicinal chemistry data from diverse public resources is crucial for academic drug discovery and translational research.
• Discrepancies in compound representation (e.g., tautomers, racemates) across databases necessitate organized hierarchies for effective bioactivity data analysis.
• Existing pipelines lack comprehensive standardization, leading to data inconsistencies and potential loss of valuable information.

Purpose of the Study:

• To develop and implement canSARchem, a novel compound registration and standardization pipeline within the canSAR public knowledgebase.
• To create compound hierarchies that facilitate the exploration of bioactivity data by organizing related chemical structures.
• To evaluate the strengths and limitations of using compound hierarchies for bioactivity data exploration and identify benefits for drug discovery.

Main Methods:

• Developed a five-step pipeline using KNIME for compound registration and hierarchy generation: structure checking, standardization, canonical tautomer/representative structure generation, salt stripping, and abstract structure generation.
• Implemented compound canonicalization prior to parent structure retrieval, similar to PubChem's approach, differentiating it from ChEMBL's pipeline.
• Assessed the impact of compound hierarchies on bioactivity data exploration and compared canSARchem's performance with existing PubChem and ChEMBL pipelines.

Main Results:

• canSARchem demonstrates a lower compound rejection rate compared to both PubChem and ChEMBL pipelines.
• FDA-approved drugs exhibit statistically significant sensitivity to canonicalization, highlighting its importance in drug discovery.
• The pipeline successfully standardized over 3 million compounds in canSAR, enabling efficient data integration and identification of alternative compound forms.

Conclusions:

• canSARchem provides an efficient method for standardizing chemical compounds and creating hierarchies, enhancing bioactivity data exploration.
• The pipeline shows comparable performance to PubChem and ChEMBL in standardization, with unique capabilities in tautomer canonicalization and a lower rejection rate.
• Compound hierarchies are essential for effective bioactivity data exploration, facilitating the identification of relevant information for drug discovery and translational research.